Mapping of two "neutralizing" epitopes of a snake curaremimetic toxin

Mapping of two "neutralizing" epitopes of a snake curaremimetic toxin by proton nuclear magnetic resonance spectroscopy. Sophie Zinn-Justin, Christian...
1 downloads 0 Views 895KB Size
Biochemistry 1993, 32, 6884-6891

Mapping of Two “Neutralizing” Epitopes of a Snake Curaremimetic Toxin by Proton Nuclear Magnetic Resonance Spectroscopy Sophie Zinn-Justin, Christian Roumestand, Pascal Drevet, Andre MBnez, and Flavio Toma’ DPpartement d’lngsnierie et d’Etude des Protsines, CE-Saclay, 91 191 Gif-sur- Yvette Cedex, France Received December 28, 1992; Revised Manuscript Received April 1 , 1993

ABSTRACT: Two monoclonal antibodies, called M a l and Ma2,3, have been previously shown to neutralize the toxic activity of the curaremimetic toxin a from Naja nigricollis. In this paper, we report the mapping of the two corresponding epitopes, using affinity chromatography and proton 2D-NMR spectroscopy. The H-D exchange rates of labile amide hydrogens have been measured in toxin a bound to each antibody and in toxin a alone. Analysis of the exchange data revealed two regions containing amide hydrogens with decreased exchange rates in the bound toxin compared to the free toxin. These two regions correspond to the sites of interaction with M a l and Ma2,3, respectively. They are consistent with prior biochemical mapping studies, and they include several residues that were not previously identified. Thus, the two antigenic sites are found to be centered on two different loops of toxin a. Comparison of these antigenic sites with the active site of toxin a allows us to delineate the molecular events associated with the two neutralization processes.

The sites by which proteins interact with antibodies play a critical role in the protection process of the immune response and are therefore currently the subject of extensive studies. The complete three-dimensional description of epitopes has been achieved in a few cases by X-ray diffraction [for reviews, see Davies et al. (1990) and Tulip et al. (1992)l. Thus, it has been shown that epitopes are constituted of about 15 residues located in several discrete segments of the polypeptide chain; the segments form a contiguous interaction surface as a result of the folding of the protein. This topographical property of epitopes makes their identification difficult and hence limits the number of mapping methods. Several approaches have been developed which avoid going through crystallization of the antibody-antigen complex. However, these methods only provide a partial view of the antigenic site. They include the use of panels of evolutionarily related proteins to identify immunodominant residues (Urbanski & Margoliash, 1977; Jemmerson & Margoliash, 1979; Benjamin et al., 1984), analysis of the immunoprotection of antigenic residues against chemical modification (Burnens et al., 1987) or proteolysis (Jemmerson & Paterson, 1986a,b), and mapping with synthetic peptides (Benjamin et al., 1984; Berzofsky, 1985; Jemmerson & Paterson, 1986a,b). Chemical modifications (Jemmerson & Margoliash, 1979,1981; Boulain et al., 1982; Trdmeau et al., 1986; Cooper et al., 1987; Collawn et al., 1988) and site-directed mutagenesis of the protein antigen (Pillet, 1991) have also been used to identify residues involved in the antibody-antigen interactions. Recently, it has been shown that proton 2D-NMR’ provides a means for the mapping of protein epitopes (Paterson et al., 1990; Paterson, 1992; Mayne et al., 1992). The principle of the method, which has also been used to study other proteinprotein complexes (Brandt & Woodward, 1987; Werner & Wemmer, 1992), is the following. Exchange behavior can be observed by NMR for a substantial proportion of individual amide and labile side-chain hydrogens; the effect of antibody binding is characterized by comparing the exchange rates

* To whom correspondence should be addressed.

Abbreviations: NMR, nuclear magnetic resonance; 2D, twodimensional; COSY, correlated spectroscopy.

0006-2960/93/0432-6884$04.00/0

measured on the free protein and on the protein bound to the antibody: the protons that become protected against exchange by antibody shielding define the binding region of the antibody in the protein (Paterson et al., 1990). The purpose of the present work is to determine and compare by this approach the interaction sites of a curaremimetic toxin with two neutralizing antibodies. The antigen, toxin a from Naja nigricollis venom (Eaker & Porath, 1967), is a structurally well-characterized protein (Zinn-Justin et al., 1992), which binds specifically and with high affinity to the nicotinic acetylcholine receptor, thus blocking the n e r v e muscle transmission. The two neutralizing antibodies, called M a 1 and Ma2,3, inhibit the binding of toxin a to the nicotinic acetylcholine receptor (Boulain et al., 1982; Trdmeau et al., 1986). They seem to act in different ways, since Ma1 accelerates the dissociation of the toxin a-receptor complex (Boulain & Mbnez, 1982), whereas Ma2,3 has no such effect on the dissociation rate of the complex (our unpublished data). Understanding of the molecular events that are associated with the neutralization processes requires the comparison and hence the delineation of both the toxic site and the epitopes recognized by M a l and Ma2,3. Previous attempts in this respect have been made by measuring the receptor binding affinity of various toxin a analogs, including chemically modified derivatives of toxin a [reviewed in Endo and Tamiya (1987, 1991)] and site-directed mutants of the highly homologous curaremimetic erabutoxin a (Pillet et al., 1993). Thus, it has been shown that the curaremimetic site essentially involves residues in loops 2 and 3 of short-chain curaremimetic toxins. Conversely, only a few residues of the M a l or Ma2,3 epitopes have been identified (Boulain et al., 1982; Trdmeau et al., 1986); therefore, additional evidence is required for the complete characterization of the antigenic sites. In this paper, we show that the interaction of toxin a with the M a l and Ma2,3 antibodies affects the amide exchange behavior in two topographically distinct regions of the toxin. From these NMR data and the previous biochemical results, the two binding sites are described. Finally, the corresponding neutralization processes are discussed. 0 1993 American Chemical Society

Mapping Two Snake Toxin Epitopes by NMR

Biochemistry, Vol. 32, No. 27, 1993 6885

loop 2 FIGURE1: Stereoview of the average backbone structureof toxin a (Zinn-Justin et al., 1992). The black spheres represent the amides of toxin

a whose exchange kinetics can be

followed in the complex by the indirect method, provided that these amides do not fully exchange in the complex before the recording of the first spectrum. They correspond to protons which do not fully exchange between the end of the exchange time in the complex and the beginning of the recording of the NMR signal, Le., to protons with a kinetic constant (kat.) lower than 1/60m i d at pH 3.5 and 283 K. MATERIALS AND METHODS Hydrogen Exchange of Antibody-Bound Toxin a. One hundred milligrams of monoclonal antibody was coupled to Materials 20 mL of Hydrazide Avidgel (Interchim); the Hydrazide Toxin a was purified from Naja nigricollis venom (Institut Avidgel allowed oriented FC region specific attachment of Pasteur, Paris, France) as previously published (Fryklund & antibodies to the gel Phase. Eight milligrams oftoxin a was Eaker, 1975). Preparation of the monoclonalantibodies M a l loadedontotheantibody Co~Umn.Unbound toxin was removed and Ma2,3 (IgG2a subclass) has been described by Boulain by COPiOUSlY washing the column with potassium Phosphate et al. (1982) and by Tr6meau et al. (1986), respectively. The buffer (50 mM KH2/K2HP04 at PH 7.4). About 7 mg of immunization was achieved by injecting increasing doses of antigen remained fixed to the antibodies. In order to initiate exchange, the column was rapidly washed toxin a into mice. The fusion procedure was carried out (1 mL/min) with 3 dead volumes of 50 mM KD2/KzDP04 according to KBhler and Milstein (1975). The resulting monoclonal hybridoma cells which produced antitoxin a were at pH 7.4 and was incubated at 283 K during periods of time ranging from 1 to 24 h. Ten minutes before the end of the grown intraperitoneally in mice. Antibodies were purified from ascitic fluids by affinity chromatography on a toxin incubation, thecolumn was washed with a low-capacity buffer (1 mM KDz/KzDP04 at pH 7.4). The toxin was then eluted a-sepharose column. Deuterium oxide, DC1, NaOD, and deuteriated acetic acid in about 10 mL of 0.2 M acetic acid at pH 2.5 and 277 K and was immediately lyophilized. The fraction was redissolved in were from CEA (99.98 atom 9% D). Fifty millimolar K2D/ KD2PO4, pH 7.4, was obtained by dissolving the salt in DzO, 1 mL of DzO, its pH was eventually readjusted to 3.5 with lyophilizing the solution, redissolving the salt in D20, and microliter amounts of DCl or NaOD, and the fraction was adjusting the pH to 7.4 with microliter amounts of dilute DC1 again immediately lyophilized. The procedure was repeated for seven different exchange times: 1, 2, 3, 5, 8, 16, and 24 or NaOD. The pH of deuteriated solutions was measured h. The same column was used for the seven experiments. without correction of the isotopic effect. Finally, each sample was analyzed by 2D-NMR Methods spectroscopy: the fraction was redissolved in DzO, and an Hydrogen Exchange Of Free Toxin a. Proton4xterium absolute value COSY spectrum was immediately recorded at exchange ofamide groups was Studied on a 7 mM Sample of 283 K on the Bruker AMX600 spectrometer. The experiment consisted of 64 scans of 1024 complex data points covering toxin a lyophilized from HzO and redissolved in D20, pH 3.5. The exchange kinetics of the individual protons was followed a spectral width of 7812.5 HZ for 256 tl increments. Total with time at 283 K by 2D-NMR absolute-value COSY acquisition time was 4 h, 11 min, experiments(Aueet al., 1976;Nagayamaetal., 1980)recorded Peak volumes were measured for the resolved (NH,CaH) On a Bruker spectrometer' Each experiment cross peaks, and the exchange rates (khund) were determined consisted of eight Scans of 1OZ4complex data points covering from their exponentialdecay as a functionofthe H-D exchange time. A protection factor, = kfrw/kbund, was finally a spectral width of 7812.5 Hz for 256 t1 increments. Total acquisition time was 25 min, so that spectra could be recorded calculated for each amide in both complexes. every 30 min during 24 h. Exchange rates (koh) were determined by a least-squares fit of the exponential decay of RESULTS the (NH,CaH) peak volumes as a function of time. The same method was applied to calculate proton4euterium Frame of the Study. As shown by the previous 2D-NMR study of toxin a (Zinn-Justin et al., 1992), the (NH,CaH) exchange rates (kfrcc)on a sample of toxin a redissolved in cross peaks corresponding to most amide protons are well 50 mM K2D/KD2PO4, pH 7.4, at 283 K. ~

6886 Biochemistry, Vol. 32, No. 27, 1993 separated; therefore, the individual exchange kinetics could be followed in free toxin a by recording the 2D-NMR spectra directly and continuously while exchange was in progress. Since the first NMR experiment was started a few minutes after the toxin was dissolved in DzO, the kinetic constants of relatively fast exchanging protons (Le., l / k > 15 min) could be measured. The same experiments were not feasible directly on the antibody-toxin complex. Therefore, an indirect approach was used to study the exchange in the toxin bound to the antibody: after exchange had occurred in the complex at pH 7.4 for different periods of time, spectra corresponding to the different exchange times were recorded on toxin a alone at low pH in order to minimize further exchange. The two approaches have been shown to yield comparable kinetic constants (Werner & Wemmer, 1992). However, theindirect method does not allow the observation of protons that fully exchange during the time period corresponding to the elution of the toxin from the complex and the preparation of the NMR sample. Thus, in this method, the observable protons are limited to those of the amides that are not fully exchanged after 1 h at pH 3.5 and 283 K. Analysis of the exchange data at pH 3.5 and 283 K showed that 32 amides have a kinetic constant (kOb)such that l/kob > 60 min. As illustrated in Figure 1, the 32 amides are widely distributed throughout the whole toxin; thus, information on the epitopes is potentially accessible in the entire structure. More precisely, eight of these amides are in loop 1, fourteen are in loop 2, five are in loop 3, and five are in the C-terminal part of the molecule. They are predominantly found in the @-sheetregions of the toxin and are rarely found at positions i, i 1, and i 2 of turns, Le., at the tips of the loops of the toxin. It must be stressed that none are found at the tip of loop 2. In the following, we will only refer to these 32 amides which can be observed by the indirect method. Exchange Measurements in the Free Toxin. The H-D exchange rates in free toxin a have been measured at pH 7.4 and 283 K, in order to further compare them to the exchange rates measured under the same conditions in the toxin bound to the antibody. Among the 32 potentially observable amides, 19 are still observable after 15 min of exchange at pH 7.4 and 283 K. However, exchange rates could be measured only for 12 of them (Table I). The seven remaining amides, corresponding to C3, H4, C23, Y24, K26, E37, and C54, are totally unexchanged during the longest incubation time: obviously, they do not constitute appropriate probes for the investigation of a H-D exchange slowing associatedwith complexformation. Only an increase of their exchange rates might be observed upon complex formation and thus might provide structural and dynamics information on the complex. Exchange Measurements in the Antibody-Toxin Complexes. Figure 2 presents COSY spectra of toxin a recorded after incubation with M a 1 and Ma2,3 over two different times. Each spectrum shows numerous cross peaks that are not observed in the spectrum of the free toxin recorded at the same exchange time. These additional peaks result from antibody binding. They correspond to different residues according to the antibody which binds toxin a. The curves in Figure 3 illustrate more quantitatively the difference of exchange behavior between the amides in the Mal-bound toxin and the amides in the Ma2,3-bound toxin. Thus, the exchange of the amide of N5 is slowed when the toxin binds to M a l , whereas it is not affected when the toxin binds to Ma2,3. In contrast, the amide of R38 exchanges more slowly when the toxin is bound to Ma2,3 compared to the toxin in the free state or when bound to M a l .

+

+

Zinn-Justin et al. Table I: H-D Exchange Rates of the Amide Protons in Free Toxin a (kfm), in Toxin a Bound to Mal (kl), and in Toxin a Bound to Ma2.3 (kz.3)" residue l/kfm (min) llk2.3 (min) 72,s Ilk1 (min) TI >92 120 1 >10000 N5 110 30 4300 143 lo0 K25 200 3060 15 1120 6 V27 50 1540 31 60 1 W28 2980 >10000 >3 7550 3 135b 97 70 >5 136b 35 70 >5 R38 30 1740 58 70 2 G41b 7